I was thrilled to be invited to the inaugural Evident AI Symposium in New York City, 7 hours of deep conversation with banking leaders driving AI in their organizations, from organizations like JPM, RBC, HSBC, Deutsche Bank, Citi and more.

The backdrop for the event was the publication of the second Evident AI Index, an independent benchmark for commercial AI adoption and performance.

This was a stellar event – it is not often one receives a real-time update on how major North American and European banks are deploying artificial intelligence, truly a day well spent.

A few key points really resonated with what I have been observing and discussing across other industries; I’m happy to discuss if these can help shape your approach to generative AI:

·       Traditional AI has already been in play in the banking industry for 10+ years.

·       While tangible financial ROI is still an aspiration for many, JPM (ranked #1 in the Index) shared that they measure, and will achieve, financial ROI from traditional AI this year. To get to this point has been a 10-year AI journey.

·       Whilst all the banks who participated are in extensive experimentation with generative AI, only Morgan Stanley spoke about generative AI being in production use, thanks to an early agreement with OpenAI to develop an internal facing AI chatbot for their financial advisors. They piloted for 9 months before moving to production, and the big lift was getting the high level of data quality needed to provide reliable results (read more below on pragmatic advice).

·       The banks sharing their experience stated that they expect to move to production use of generative AI anywhere from 3 months to 1+ years from now, with most products likely internal-facing.

·       Data was a constant topic in its quality, fairness, reliability, and commercial value.

·       Another constant topic: people. All attending organizations foresee AI augmenting, not replacing, their workforce. Education on generative AI and involvement from within their organizations is key, whether employee driven use-case innovation or the nomination of AI Ambassadors across their firms.

·       Technology roadmaps and investments are being evaluated/re-evaluated given the accelerated pace of generative AI. Of note, this doesn’t mean more dollars being allocated to accommodate generative AI, rather a refocus of budget dollars.

·       Most of the banks will have multi-model strategies, with few stating the intent to build their own large language models, though this could change. Everyone agreed that it is difficult to predict what the situation will be 3-5 years from now. In fact, everything being discussed now can change in 6 months.

·       The most pragmatic advice of the day was from the Head of Analytics, Data & Innovation for Morgan Stanley Wealth Management who said to “just do something [with generative AI] …really there are no experts so get the smartest people [you have] in the room”. He also noted that their impetus for generative AI use is to address business problems and provide value to their organization and clients. However, their first production initiative was focused on learning how to leverage generative AI rather than achieving financial KPIs. Through this initiative, they not only learned how to effectively deploy generative AI, but also that the chatbot assistant has demonstrated business value.

·       The general sentiment was that experimentation with generative AI is essential, whilst production use must be aligned with business strategy and drive value, in the context of acceptable risk and regulation.